Multiple nodes, what is the best configuration?

Good morning, I am new to the forum, I have several questions like all noob, how many nodes can I have without receiving a penalty from the system (same storj address, same email) (all nodes with different public IP).
I have 3 days with 2 nodes mounted ( 2TB storage for the moment) and the interface reads Disk Space Used This Month 1.08TB*h, what does that mean?
Which suits me best, 2 nodes with 8TB each, or 8 nodes with 2 TB each (all nodes with different public IP over RAID5 y Ceph in two servers Xeon)?

As i know, there are no penalties. You can use one e-mail to multiple nodes, as well as wallets. You can even have them on the same IP, but they will share the amount of receiving data.
RAID5 is not recommended, better use RAID6 in this case or single node per hdd.
Xeon is good, but not needed for this task, you will probably will have a bigger power consumption.

It’s a wrong assumption: Node Operator Terms & Conditions

och… ok. thank you, this was my mistake.

Operate more than one (1) Storage Node behind the same IP address;

obviously we can’t really fully rely on that link if it is still outdated.
Running more than 1 node behind the same IP address is common practice since a long time.

4 Likes

another “restriction” and or limitation is that you will not get more data with more nodes, the data assigned to your global ip segment will be distributed amongst the nodes in that segment… which would in most cases be your local network nodes only…

this can be used for several things… like load balancing when using limited hardware such a SMR, even tho i would prefer hdd redundancy instead of multiple nodes.

however with stuff like SMR drives multiple nodes are most likely the best option due to the inherent write speed limitations of the technology.

however keep in mind… hdd’s are basically the same basic concept as vinyl records… just using magnetics and metal disks… they are quite unreliable for long term high capacity storage.

so raid will make your life simpler in some ways, but ofc at the cost of profit, if you can trust your hdd’s to store the data correctly…

ofc raid can be painful to manage at times, and most certainly you shouldn’t use raid5
you will be better off using individual disks / nodes on each hdd rather than using raid5

raid5 creates as many problems as it solves… ofc it is slightly more reliable than single drives… atleast when first installed… but in its later stages of life… often you end up having multiple drives that are failing and you aren’t aware until you want to replace the catastrophically failed drive… there is also the problem with raid 5 that it cannot tell which disk is right or wrong…

if you get an error on a drive… then it has a parity data… but how does it know if the parity is wrong or the stored data is wrong… it can’ts it’s basically a 50/50 shot… then it will check smart and see if the disk has reported problems and then the one that has smart issues will be the one it assumes is wrong… thus raid5 can have a 1 byte error and still contain the correct data in the array only to then delete it … without ever asking you.

so yeah don’t use raid 5… used raid6 or mirrors, don’t think we have anyone here actually using mirrors for their raid setup… but better than raid 5 atleast … but again has the same issue… it cannot see which drive is wrong and which is right…

in raid6 you have 3 entries pr of redundancy… thus you can have a vote basically… if you have a bad bit of data… you compare all the blocks and the data that is still stored in two locations is what you go with… that way raid6 can know with atleast 99% certain or better that it has the correct data, before it writes the correction to the bad drive.

or rebuilds a new hdd put into the system.

ofc sometime raid makes everything much harder… not always fun to try to migrate a node on a raid… but i like raid… especially now that i really understand how unreliable hdd’s really are